No luck. I guess I just have to come to terms with the fact that the idea that new Raspbian versions are the definitive, up to date solution for old models is just a lie. The hardware decoding for the 3B+ is just broken. Just surprised how little I see about it.

This may seem pedantic, but mp4 is a container that holds the video and audio streams. The actual video stream can be encoded im various formats (mpeg 2, h264, h265, etc). If you open vlc and look at the codec menu, find the video stream and report back the encoding type it may provide some insight. It could be that there's a performance regression with a particular decoder or maybe they changed decoding library or any number or things. Sorry it's a bit vague, but what I'm getting at is if we know the actual video encoding of the file it may help to track down the decoding performance issue.

If it does turn out to be mpeg2, it could be that something changed about how the video decoding drivers (kernel module) are loaded. Like maybe they stopped including them by default or are no longer being used for some reason.

If it's not mpeg 2, then could look in to decoder specific changes between distro versions or hardware support related changes (like maybe a kernel module needs an extra config passed to it to get better performance on 3b), or even decoder library config may be possible to tweak. Sometimes performance optimizations make things worse and the new default configs work better on newer hardware but worse for you.

In any case, I think knowing the specific video encodings would be helpful. I also just remembered that I had some performance issues on some files due to audio formats if I was having the Pi software decode vs connected to an external AV receiver that could decode the bit streamed audio data.